Entomological knowledge in Madagascar by GBIF datasets: estimates on the coverage and possible biases (Insecta)
Keywords:Madagascar, Insect distribution, Global Biodiversity Information Facility (GBIF), Vegetation macrogroups, GIS analysis, Data coverage, Protected areas system, Hotspot analysis
Although Madagascar is one of the world’s most important biodiversity hotspots, the knowledge of its faunistic diversity is still incomplete, notwithstanding many field campaigns were organized since the 17th century until nowadays, leading to a huge number of vertebrate and invertebrate records. In this contribution, taking into consideration the geographic distribution by a GBIF dataset including 286,764 records referred to nine insect orders (Coleoptera, Diptera, Hemiptera, Hymenoptera, Lepidoptera, Neuroptera, Odonata, Orthoptera, Trichoptera), we tried to supply some observations on the spatial distribution and to point out some possible biases in the entomological knowledge of Madagascar. Hymenoptera, Coleoptera and Diptera were the most represented orders in the dataset, respectively. Some orders show many “coupled” sampling, with peaks of shared sampled localities between Diptera with Hymenoptera (98.07%) and Hemiptera with Coleoptera (64.21%). Considering the geographic location and the extension of the vegetation macrogroups in Madagascar, the entomological data result unevenly distributed. Current Protected Areas’ (PAs) network covers about the 70% of the total of the collecting localities for the nine insect orders considered, even though some, such as Trichoptera, Odonata, and Neuroptera seem significantly less protected than others. However, the possible new PAs planned for Madagascar could greatly increase in the future the protection level for all 9 insect orders analyzed, especially for Neuroptera, Odonata and Lepidoptera. A percentage of 82.3% of the whole sampling localities falls inside the PAs themselves or within 1000 m from their borders. A similar pattern is observed for the road network: the 62.9% of the localities fall at least at 1000 m from a road, with no sampling localities observed further than 10 km from a road; statistically significant clusters were observed in evaluating these biases, coinciding with major towns or PAs.